Fingerprint analysis is amongst the most widely used and studied biometric techniques. During the last two decades, many new and exciting developments have taken place in the field of fingerprint science, summarized for example in the monograph Advances in Fingerprint Technology, 2nd ed., edited by H. C. Lee and R. E. Gaensslen (CRC Press, 2001). Fingerprint identification not only plays a major role in forensic or police science, but also in controlling the building-access or information-access of individuals to buildings, rooms, and devices such as computer terminals.
Typically in electronic fingerprint matching, a live fingerprint is scanned and electronically digitized. The digitized data generally contains information pertaining to characteristic features of the fingerprint, such as ridge endings, points of ridge bifurcation, and the core of a whorl, i.e. fingerprint minutiae. The digitized data is then compared with stored data relating to fingerprints that have been obtained previously from corresponding authorized persons, i.e. fingerprint templates. When a match is detected, within a predetermined level of security in the form of a predetermined false acceptance rate, the individual is identified and a corresponding action is performed.
In general, there are two types of errors associated with fingerprint identification. The first is false reject or Type I error, and the second is false accept or Type II error. A Type II error occurs when there are enough similarities between fingerprints of two individuals, that one is mistaken for the other. A Type I error occurs for a variety of reasons, and refers to when an individual is not identified even though the individual is an authorized user registered with the system.
It has been suggested that the underlying cause of errors in fingerprint analysis is that the amount of data from a fingerprint is too limited for it to be used in a biometric identification system involving a large number of users. Typically, there are only 30 to 40 minutia points available in a fingerprint. Alternatively, poor performance of fingerprint verification systems is attributed to the fact that the final verification decisions are based upon comparisons of small isolated regions of the fingerprint, when in fact a small sampling of the ridges does not provide enough detail to accurately verify or identify an individual.
One way to increase the accuracy of fingerprint identification is to include the analysis of pores within the ridges of the fingers, such as sweat pores. Pores are naturally occurring physical characteristics of the skin, which have conventionally been ignored in biometric identification. However, a typical finger contains about 50 to 300 pores, each of which varies in size and shape, and that is used in combination with other pores and pore locations for uniquely identifying an individual. Furthermore, the analysis of pore prints generally obviates the fraudulent problems encountered with fingerprint identification, since latent fingerprints do not generally contain a lot of detail about pore size, shape and/or distribution. Consequently, the use of latent fingerprints, which often fools most fingerprint identification systems, does not deceive pore print identification systems.
In a paper entitled Automated System for Fingerprint Authentication (Proc. SPIE 1994, 2277, 210-223), Stosz et al. describe a novel technique for automated fingerprint authentication, which utilizes pore information extracted from live scanned images. The position of the pores on the fingerprint ridges is known to provide information that is unique to an individual and is sufficient for use in identification. By combining the use of ridge and pore features, a unique multilevel verification/identification technique has been developed that possesses advantages over systems employing ridge information only. An optical/electronic sensor capable of providing a high resolution fingerprint image is required for extraction of pertinent pore information, which makes it unlikely that electronically scanned inked fingerprints would contain adequate pore data that is sufficient, or consistent enough, for use in authentication. The feasibility of this technique has been demonstrated by a working system that was designed to provide secure access to a computer. Low Type I error rates, and no Type II errors have been observed based on initial testing of the prototype verification system. It has been suggested by Stosz et al. that a high-resolution scanner of at least 800 DPI or greater is required to accurately resolve pores.
U.S. Pat. No. 5,982,914 to Lee et al. issued Nov. 9, 1999, discloses a method of identifying an individual using an analysis of both pore locations and minutia data. The method comprises obtaining from an individual during a registration process, a fingerprint image having at least one registration pore and at least one registration macrofeature, that is a ridge or minutia. In a bid step, a fingerprint image having at least one bid pore and at least one bid macrofeature is obtained. Bid associated data is compared to the registration associated data to produce a correlation score, wherein a successful or failed identification is obtained based on comparison of the correlation score to a predetermined threshold value. In the teachings of Lee et al. the analyses were accomplished using a commercially available 500 DPI resolution scanner.
Although the prior art clearly benefits from pore print technology, it does not truly exploit the advantages of pore print identification. Specifically, in the system disclosed by Lee et al. the commercially available 500 DPI resolution scanner is not able to extract pore detail relating to the size and shape of the pores. In fact, the teachings ignore the detailed pore morphology and concentrate on pore distribution with respect to fingerprint minutiae. As a result, this system incurs many of the same limitations as conventional fingerprint identification systems. Specifically, the individual must have an undamaged fingerprint, the imaging device must be able to acquire a full fingerprint image (i.e., a surface area of approximately 3×2 cm2 size), and the individual must precisely align the predetermined finger or thumb in a manner that allows a proper matching. In the system disclosed by Stosz et al. the high-resolution scanner, which is large, bulky, expensive, and difficult to manufacture, is the greatest disadvantage,
It has now been found that it is only necessary to scan a small sampling area of a live pore print for comparison to a portion of the stored pore print template, since pores, unlike fingerprint minutiae, are distinguishable from much smaller cross-sections. For example, if there are typically only 30 to 40 minutia points available from a fingerprint covering an area of approximately 2×1 cm2, then an imaging device with a scanning area one quarter of that size statistically captures a quarter of the minutia points. Clearly, it would be highly inaccurate to base a biometric identification system on such a small amount of information. Furthermore, it is highly unlikely that the characteristic minutiae, for example the core of the whirl, would be captured with a small sensor due to inconsistent sampling methods. In pore print analysis, the size, shape and location of the pores all contribute detailed information, thus increasing the accuracy of the biometric identification system and allowing a smaller scanning area to be used. In addition, the large number of pores typically available in a small cross-section of skin further contributes to the accuracy when sampling a smaller cross section. The difficulties in fingerprint analysis associated with inconsistent sampling methods are not a problem in pore print analysis, since any portion of the pore print is used to characterize the individual i.e., there is no need to capture localized minutiae as required in fingerprint analysis. Often, pore print registration provides a suitable cross-section of pores for characterization on a first attempt.
However, when only small regions of a fingerprint are sensed, difficulties of a different nature arise. Although the partial pore print provides sufficient information for identifying an individual by establishing that the scanned fingerprint is identical to one of the template fingerprints, this identification process is costly in processing time and processing resources, due to the high number of permutations needed when comparing the sensed profile with partial master profiles. The high number of permutations also adds a degree of uncertainty to the identification process. As a consequence, the type II error rate increases, requiring a less stringent predetermined false acceptance rate. It would be advantageous to have supporting information available, assisting in the step of selecting partial pore print patterns from the template database, which are likely candidates to provide a positive match with the scanned pore print profile.
Also, when scanning a series of partial areas of the fingerprint, a method that is described in U.S. Pat. No. 6,333,989 to Borza issued Dec. 25, 2001, it is important that the finger moves in a defined motion across a sensing pad. When the sensing pad is constructed such that it requires a fingertip to move in one particular direction, deviations from the optimal scan direction reduce the amount of information gathered during the scanning process, and complicate the identification process.
It would be advantageous to have a device at one's disposal, which, although utilizing the technique of partial area scanning, does not posses an inherent directional dependence on the direction of movement of the fingerprint to be imaged.
It would be of further advantage when the partial images recorded during a fingerprint scan cover the area of a fingerprint as complete as possible, so as to provide a maximum of information, as to establish the use of low false acceptance rates.
It would also be advantageous to provide a device that is small, robust, and cost efficient, and at the same time allowing to record high resolution images.